The risk linked to ionizing radiation: an alternative epidemiologic approach.

Radioprotection norms have been based on risk models that have evolved over time. These models show relationships between exposure and observed effects. There is a high level of uncertainty regarding lower doses. Recommendations have been based on the conservative hypothesis of a linear relationship without threshold value. This relationship is still debated, and the diverse observations do not allow any definitive conclusion. Available data are contradictory, and various interpretations can be made. Here we review an alternative approach for defining causation and reconciling apparently contradictory conclusions. This alternative epidemiologic approach is based on causal groups: Each component of a causal group is necessary but not sufficient for causality. Many groups may be involved in causality. Thus, ionizing radiation may be a component of one or several causal groups. This formalization reconciles heterogeneous observations but implies searching for the interactions between components, mostly between critical components of a causal profile, and, for instance, the reasons why specific human groups would not show any effect despite exposure, when an effect would be expected.